By defining the 8-seed game as the best champions outside the top six, and the 7-seed game as the best at-large teams outside the top six, it's possible for the 8-seed game to have "better" teams than the 7-seed game. You don't want the 2-seed to be better than the 1-seed.

You could fix that problem by seeding the last four together for the play-in (7 vs 10, 8 vs 9), but that would reduce the benefit for being a 1-seed in the case where the last two chosen are significantly worse. You could pair the two worse together (7 vs 8, 9 vs 10), but now the 9 seed is given an advantage over the 7 seed in winning their first game, without respect to conference championships. Probably the best solution would be to reseed depending on the winners of the play-in games. Trying to adopt the play-in methods of the NCAA basketball tournament won't work because your playoff is too small to ensure teams "deserve" play-in status.

If all teams happened to be in the top 10, then the benefit of top seeding is similarly reduced, there's not much difference between facing the 5th best and the 8th-10th best in the initial game. You never want to create a situation where the probable #1 seed has an incentive to lose its last game for matchup purposes.

I prefer sixteen with autobids for every conference. The conference champions your system is trying to exclude give more compelling reasons for top teams to care about seeding, and also adds interest to the conference race in *every* conference. It also creates six more football games to sell without extending the length of the postseason. To my mind, the ideal postseason format is the one that maximizes regular season interest. I think this is best accomplished with all conference champions and *very limited* at large bids. I think the NCAA basketball tournament is too big, not because of the cannon fodder one-bid conferences, but because the large number of at-large bids and the conference tournaments take a lot of the meaning out of regular season play.